Created fallback for lob path

Author: AntonioG70

dbptk-modules/dbptk-module-siard/src/main/java/com/databasepreservation/modules/siard/in/content/SIARD2ContentImportStrategy.java

@@ -103,6 +103,7 @@ public class SIARD2ContentImportStrategy extends DefaultHandler implements Conte
   private int currentColumnIndex;
   private Row row;
   private long rowIndex;
+  private boolean useLobPathFallback = false;
 
   public SIARD2ContentImportStrategy(ReadStrategy readStrategy, ContentPathImportStrategy contentPathStrategy,
     SIARDArchiveContainer lobContainer, boolean ignoreLobs) {
@@ -322,16 +323,16 @@ public void startElement(String uri, String localName, String qName, Attributes
                   new DummyInputStreamProvider(), lobPath, optionalLength.orElse(0L), optionalDigest.orElse(null),
                   optionalDigestType.orElse(null));
               } else {
+                inputStream = createInputStream(container, lobPath, lobDir);
                 currentBlobCell = new BinaryCell(
-                  currentTable.getColumns().get(currentColumnIndex - 1).getId() + "." + rowIndex,
-                  readStrategy.createInputStream(container, lobPath));
+                  currentTable.getColumns().get(currentColumnIndex - 1).getId() + "." + rowIndex, inputStream);
               }
             }
 
             LOGGER.debug(
               String.format("BLOB cell %s on row #%d with lob dir %s", currentBlobCell.getId(), rowIndex, lobDir));
           } else if (lobDir.endsWith(SIARD2ContentPathExportStrategy.CLOB_EXTENSION)) {
-            inputStream = readStrategy.createInputStream(container, lobPath);
+            inputStream = createInputStream(container, lobPath, lobDir);
             String data = IOUtils.toString(inputStream);
             currentClobCell = new SimpleCell(
               currentTable.getColumns().get(currentColumnIndex - 1).getId() + "." + rowIndex, data);
@@ -474,6 +475,22 @@ public void characters(char buf[], int offset, int len) {
     tempVal.append(buf, offset, len);
   }
 
+  private InputStream createInputStream(SIARDArchiveContainer container, String lobPath, String lobDir) throws ModuleException {
+    String lobName = Paths.get(lobDir).getFileName().toString();
+    if (useLobPathFallback) {
+      lobPath = contentPathStrategy.getLobPathFallback(null,
+        currentTable.getColumns().get(currentColumnIndex - 1).getId(), lobName);
+    }
+    try {
+      return readStrategy.createInputStream(container, lobPath);
+    } catch (ModuleException e) {
+      useLobPathFallback = true;
+      lobPath = contentPathStrategy.getLobPathFallback(null,
+        currentTable.getColumns().get(currentColumnIndex - 1).getId(), lobName);
+      return readStrategy.createInputStream(container, lobPath);
+    }
+  }
+
   private List<Integer> getArrayCellPositions(Integer current, Deque<String> parentTags) {
     List<Integer> positions = new ArrayList<>();
 

dbptk-modules/dbptk-module-siard/src/main/java/com/databasepreservation/modules/siard/in/path/ContentPathImportStrategy.java

@@ -15,6 +15,8 @@
 public interface ContentPathImportStrategy {
   String getLobPath(String basePath, String schemaName, String tableId, String columnId, String lobFileName);
 
+  String getLobPathFallback(String basePath, String columnId, String lobFileName);
+
   /**
    * Stores the relationship between the schema and it's folder
    * 

dbptk-modules/dbptk-module-siard/src/main/java/com/databasepreservation/modules/siard/in/path/SIARD1ContentPathImportStrategy.java

@@ -48,6 +48,11 @@ public String getLobPath(String basePath, String schemaName, String tableId, Str
     throw new NotImplementedException("getLobPath in ContentPathImportStrategy method is not needed for SIARD1");
   }
 
+  @Override
+  public String getLobPathFallback(String basePath, String columnId, String lobFileName) {
+    throw new NotImplementedException("getLobPathFallback in ContentPathImportStrategy method is not needed for SIARD1");
+  }
+
   @Override
   public void associateSchemaWithFolder(String schemaName, String schemaFolder) {
     schemaFolders.put(schemaName, schemaFolder);

dbptk-modules/dbptk-module-siard/src/main/java/com/databasepreservation/modules/siard/in/path/SIARD2ContentPathImportStrategy.java

@@ -87,6 +87,27 @@ public String getLobPath(String basePath, String schemaName, String tableId, Str
     }
   }
 
+  @Override
+  public String getLobPathFallback(String basePath, String columnId, String lobFileName) {
+    if (StringUtils.isBlank(basePath)) {
+      basePath = metadataLobFolder;
+    }
+
+    String columnPart = columnFolders.get(columnId);
+
+    if (columnPart == null) {
+      columnPart = ".";
+    }
+
+    if (".".equals(basePath) && ".".equals(columnPart) && lobFileName.startsWith("..")) {
+      return lobFileName.substring(3);
+    } else if (".".equals(columnPart)) {
+      return lobFileName;
+    } else {
+      return basePath + RESOURCE_FILE_SEPARATOR + columnPart + RESOURCE_FILE_SEPARATOR + lobFileName;
+    }
+  }
+
   @Override
   public void associateSchemaWithFolder(String schemaName, String schemaFolder) {
     schemaFolders.put(schemaName, schemaFolder);

dbptk-modules/dbptk-module-siard/src/main/java/com/databasepreservation/modules/siard/in/path/SIARDDKPathImportStrategy.java

@@ -10,6 +10,7 @@
 import jakarta.xml.bind.JAXBException;
 import jakarta.xml.bind.Unmarshaller;
 import java.util.List;
+
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 import org.xml.sax.SAXException;
@@ -169,6 +170,11 @@ public String getLobPath(String basePath, String schemaName, String tableId, Str
     throw new UnsupportedOperationException("Invoking getLobPath(...) is not relevant for SIARDDK.");
   }
 
+  @Override
+  public String getLobPathFallback(String basePath, String columnId, String lobFileName) {
+    throw new UnsupportedOperationException("Invoking getLobPath(...) is not relevant for SIARDDK.");
+  }
+
   @Override
   public void associateSchemaWithFolder(String schemaName, String schemaFolder) {
     throw new UnsupportedOperationException("Invoking associateSchemaWithFolder(...) is not relevant for SIARDDK.");